Processed food and meat

In October 2015 the World Health Organization's cancer agency, IARC, classified processed meat as "carcinogenic to humans (Group 1)".¹ In 2019 a panel of fourteen people from seven countries, the Nutritional Recommendations consortium (NutriRECS), working from five systematic reviews done for the purpose, published a recommendation that sounds like the opposite: "the panel suggests adults continue current processed meat consumption (weak recommendation, low-certainty evidence)."¹³ Its own reviews had found an association running the same way IARC's evidence did. You don't need to decide that one of them misread the data. This lesson reads two words that arrive in headlines already carrying a verdict, "ultra-processed" food and processed meat, and asks what sits under each: how much is evidence, how much is argument about the evidence, and how much is a judgement about what matters.

With ultra-processed food, part of the argument is whether the category is well drawn. With meat, the numbers are less disputed than the advice, and a premise about values fills the gap.

What "ultra-processed" means

The term comes from a classification called NOVA, built by Monteiro's group. In a 2019 commentary, which this course read at abstract level, they define it: "Ultra-processed foods are defined within the NOVA classification system, which groups foods according to the extent and purpose of industrial processing."⁴ NOVA has four groups, which a critical paper lists as "unprocessed or minimally processed foods, processed culinary ingredients, processed foods, and ultra-processed foods and drinks".⁵ So "processed" and "ultra-processed" are different groups.

What makes something ultra-processed, in Monteiro's words: "Processes and ingredients used to manufacture ultra-processed foods are designed to create highly profitable (low-cost ingredients, long shelf-life, emphatic branding), convenient (ready-to-consume), hyper-palatable products liable to displace all other NOVA food groups, notably unprocessed or minimally processed foods."⁴

Notice that the definition is partly about purpose: what the product was designed to do. Critics say that's a weakness, since a nutrient can be measured in a lab and a purpose can't (this course's summary; Astrup's version is below). On this course's reading, the practical test that follows is Monteiro's answer: it turns the purpose into things you can find on a label.

For shoppers the commentary gives a practical test: "A practical way to identify an ultra-processed product is to check to see if its list of ingredients contains at least one item characteristic of the NOVA ultra-processed food group, which is to say, either food substances never or rarely used in kitchens (such as high-fructose corn syrup, hydrogenated or interesterified oils, and hydrolysed proteins), or classes of additives designed to make the final product palatable or more appealing (such as flavours, flavour enhancers, colours, emulsifiers, emulsifying salts, sweeteners, thickeners, and anti-foaming, bulking, carbonating, foaming, gelling and glazing agents)."⁴

That's a test you can run on a label, which is part of the category's appeal. The next question is whether different people applying NOVA get the same answer.

Do people agree on what's ultra-processed?

Braesco and colleagues asked French food and nutrition specialists to assign foods to NOVA groups: 120 marketed products with their ingredient lists, and 111 generic foods without. This course read the abstract.⁸ They measured agreement with Fleiss' kappa, which the abstract describes as running from 0 to 1, "where 1 = 100% agreement". In this course's gloss, it counts agreement beyond what chance alone would produce.

Hall 2019: the trial everyone cites

The trial at the centre of the argument was run by Kevin Hall's team at the US National Institutes of Health, and it's free to read. It's the same NIH team you met on a ward in lessons 1 and 3; this trial came two years before lesson 3's. This course read the abstract and, in the full text, the introduction, the diet design, the results on intake, energy density, eating rate, appetite and hormones, the limitations and the conclusion.⁹

Twenty weight-stable adults lived on a ward. Each ate an ultra-processed diet for two weeks and an unprocessed diet for two weeks, straight after, in random order. "Meals were designed to be matched for presented calories, energy density, macronutrients, sugar, sodium, and fiber. Subjects were instructed to consume as much or as little as desired."⁹ So total calories, sugar, salt and fat on the tray matched. What people chose to eat wasn't fixed.

Weight change tracked intake closely, a correlation of 0.8, against the 0.21 of lesson 2's questionnaires.⁹ People also ate faster on the ultra-processed diet, by about 17 kcal a minute ("17±1 kcal/min; p<0.0001").⁹ And the matching was less complete than the abstract's sentence suggests.

The full text lists what still differed. Added sugar was about 54% of the sugar on the ultra-processed diet and 1% on the other ("54% vs 1%"), saturated fat about 34% of the fat against 19% ("34% vs 19%"), and the insoluble share of the fibre "~16% versus 77%".⁹ Those are the differences Astrup's reading, below, leans on.

The trays were matched on energy density, the calories in each gram of food, only as presented. The authors explain that fibre supplements were dissolved in drinks served with the ultra-processed meals, and that "the ~85% higher energy density of the non-beverage foods in the ultra-processed versus unprocessed diets" (the solid food, in other words) "likely contributed to the observed excess energy intake".⁹ The authors also weighed the idea that people eat until they reach a protein target, and estimated it "could potentially explain at most ~50% of the observed energy intake differences".⁹

What it can't tell you

Two readings of the same trial

Hall's trial is read two ways, which is why it's worth reading slowly.

Hall's team drew the first reading themselves. Their abstract ends: "Limiting consumption of ultra-processed foods may be an effective strategy for obesity prevention and treatment."⁹ Monteiro and colleagues put feeding trials in general into the same case. Their fullest recent statement is a 2025 Lancet series, whose first paper this course read at abstract level. It says the harm to diet quality "is confirmed by national food intake surveys, large cohorts, and interventional studies", among them "overeating driven by high energy density, hyper-palatability, soft texture, and disrupted food matrices", and concludes: "The totality of the evidence supports the thesis that displacement of long-established dietary patterns by ultra-processed foods is a key driver of the escalating global burden of multiple diet-related chronic diseases."¹² The abstract doesn't name Hall's trial. (A food matrix is a food's physical structure, as distinct from its nutrient list; that gloss is this course's.)

Astrup argued the other side of a 2022 debate in the American Journal of Clinical Nutrition (both papers read at abstract level). Of what his abstract calls "A highly cited randomized controlled feeding study", which is Hall's, he says it "showed a rapidly weaning effect on energy intake that can be entirely explained by more conventional and quantifiable dietary factors, including energy density, intrinsic fiber, glycemic load, and added sugar."⁷ (Glycemic load is a measure of how much a food raises blood sugar, scaled by the amount eaten.)

Look at where they agree. Both name energy density. The dispute is whether "ultra-processed" is a useful name for a bundle of properties that includes it, or a detour around properties that can be measured directly. That's a question about a category, and Hall's trial wasn't built to settle it.

A longer test: Dicken 2025

In 2025 a team in England, with Hall among its authors, ran a longer trial; this course read the abstract only.¹⁰ Fifty-five adults whose usual diet was at least half ultra-processed were "provided with two 8-week ad libitum diets following the UK Eatwell Guide": one minimally processed, one ultra-processed, in random order.¹⁰ Ad libitum means eat as much as you like. The abstract does not say whether people ate at home or under supervision, and this course did not read the methods, so it can't say how free-living this was.

The design goes some way toward the critics' point about nutrients: both diets met the same national guide, so this was not a healthy diet against a poor one.

Read it at its own strength. It points the same way as Hall 2019, with a smaller effect, over a longer time, in more people, and an ultra-processed diet that met national guidance still went with weight loss from where people started. Weight here is a measure the trial chose, not a goal this course sets for anyone. As of September 2026 it is the newest trial this course read, and it is one trial.

Convincing and very low, at once

Most of the evidence on ultra-processed food and disease is observational. Lane and colleagues pooled it in a 2024 umbrella review in the BMJ, a review of 45 earlier pooled analyses covering nearly ten million people; this course read the abstract.¹¹ They graded each finding twice, on their own pre-set credibility scale and on GRADE, the system Sleep lesson 6 introduced and this course's lesson 6 used for rating how certain evidence is.

The result for deaths from cardiovascular disease: "convincing evidence (class I) supported direct associations between greater ultra-processed food exposure and higher risks of incident cardiovascular disease related mortality (risk ratio 1.50, 95% confidence interval 1.37 to 1.63; GRADE=very low)".¹¹ Convincing and very low, in one sentence, for one finding. Across all 45 analyses: "Overall, using the GRADE framework, 22 pooled analyses were rated as low quality, with 19 rated as very low quality and four rated as moderate quality."¹¹

How can one finding be both? The abstract names the classes but does not define what builds them. This course's reading is that the two scales ask different questions. A credibility class asks how strong, large and consistent an association is. GRADE asks how far the design could be misleading you, and it starts observational evidence at low. NutriRECS's patterns review, for one, says its cohort studies "provide low- or very-low-certainty evidence".¹⁷ A big, consistent association from cohorts can pass the first test and start low on the second, because consistency doesn't remove confounding.

Low is where GRADE starts cohort evidence, not where it has to stay: Lane rated four of its 45 moderate, including type 2 diabetes.¹¹ GRADE's own grounds for rating observational evidence up, as a 2020 paper by Qian and colleagues tabulates them, are a "Large magnitude of effect", a "Dose-response gradient", and confounding that "would only strengthen the observed association".²⁰ Meat can only be studied in cohorts, so whether its evidence clears those bars is what NutriRECS and its critics argue over below.

The category, for and against

For. Monteiro, arguing the "yes" side of the 2022 debate: "This recommendation is supported by systematic reviews and meta-analyses of nationally representative dietary surveys and long-term cohort studies." And, with the hedge that goes with it: "Although more research should be done to identify these mechanisms, existing evidence is sufficient to recommend the avoidance of ultra-processed foods to optimize health and policies to support and make feasible this recommendation."⁶ Their case is that surveys and cohorts link ultra-processed food with poorer diets and more disease, that official guidelines are already adopting the advice, and that the evidence is enough to act on before the mechanisms are known.

Against. Gibney and colleagues, in 2017: "On balance, therefore, there seems to be little advantage from the use of the NOVA classification compared with the current epidemiologic approach, which relies on the linkage of nutrient intakes to chronic disease with subsequent identification of foods that merit consideration in public health nutrition strategies."⁵ Astrup adds two charges: that the classification "rests on poorly defined food processes", "easily leading to misclassification", and that it "characterizes several healthy, nutrient-dense foods as unhealthy".⁷ He also says that the observational studies linking ultra-processed food with obesity have not, in his words, "demonstrated independent associations after controlling for likely confounders."⁷

Neither side claims a mechanism specific to processing is established. Monteiro says more research is needed to identify the mechanisms; Astrup says the effects are already accounted for: "little evidence supports effects beyond those that can be accounted for by nutrient composition, energy density, and food matrices".⁶⁷ What would settle the question, in this course's summary of its research, is longer feeding trials that match the two diets on energy density, eating rate and palatability as well as nutrients, to see whether any effect of processing remains, plus a classification that different people apply the same way. Dicken 2025 is a step toward the first; Braesco shows the second isn't there yet.

On interests, this course read no funding or disclosure statements for the ultra-processed papers, on either side, so it names none. Both sides do argue about policy. A companion Lancet paper by Scrinis and colleagues, Monteiro among them, sets out "government policies aimed at halting and reversing the rise of UPFs worldwide".¹² Astrup's abstract says the classification "is counterproductive to solve the major global food production challenges".⁷ A round of letters answering the 2025 series appeared in 2026, and this course didn't read those either.

Meat: what nobody disputes

Logic and Argument lesson 9 set IARC's 18% for processed meat beside Cancer Research UK's figures: about ten more cases of bowel cancer per thousand people among the heaviest eaters than among the lightest. Sleep lesson 4 taught that an IARC group grades the strength of evidence that something can cause cancer, not the size of the risk. This lesson doesn't redo either. It asks why, given those numbers, expert bodies gave different advice.

Start with what everyone who looked found. IARC's working group, "22 experts from 10 countries", who "considered more than 800 studies", put processed meat in Group 1 "based on sufficient evidence in humans that the consumption of processed meat causes colorectal cancer", and red meat in Group 2A "based on limited evidence that the consumption of red meat causes cancer in humans and strong mechanistic evidence supporting a carcinogenic effect."¹

On amounts, the 2015 Q&A said that "every 50 gram portion of processed meat eaten daily increases the risk of colorectal cancer by about 18%", and that "if the association of red meat and colorectal cancer were proven to be causal", the risk "could increase by 17% for every 100 gram portion of red meat eaten daily."² Note the "if" on red meat. And the release carried a sentence that travels less often: "At the same time, red meat has nutritional value."¹

NutriRECS's reviews, which you'll meet below, found the same direction. Nobody in this lesson's sources says the association isn't there. The disagreement begins at what it means and what to do.

A page dated 23 September 2026

WHO keeps a question-and-answer page on the 2015 classification. It carries the date 23 September 2026, and this course read the whole page the next day.³ It keeps the Group 1 and 2A wording and the 18% and 17% figures. Unlike IARC's 2015 Q&A, it doesn't carry the comparison with tobacco and asbestos or the estimates of cancer deaths a year. Asked whether people who have had colon cancer should stop eating red meat, it answers: "Meat can be part of a healthy diet and provides nutrients that are important for good health. However, many national dietary guidelines recommend limiting processed meat and red meat because higher intakes are linked to an increased risk of heart disease, type 2 diabetes and some cancers."³

Asked how much is safe to eat, it answers: "The risk increases with the amount of red or processed meat consumed, but the data available for evaluation did not permit a conclusion at the time of the original assessment about whether a safe level exists. WHO is currently developing a guideline on the optimal intake of animal-source foods, which will include recommendations on the amount of red and processed meat to consume."³ That guideline may appear while this course is live.

Three bodies, three questions

The disagreement shrinks once you see that the three bodies, IARC, the World Cancer Research Fund (WCRF) and NutriRECS, weren't asking the same question.

IARC's answer isn't advice about amounts. Its Q&A said in 2015 that the classifications "describe the strength of the scientific evidence about an agent being a cause of cancer, rather than assessing the level of risk", and the recommendation on amounts is, as of September 2026, the job of the WHO guideline still being written.²³

WCRF does give advice. Its evidence page, which this course read in full, says: "There is strong evidence that consumption of either red or processed meat are both causes of colorectal cancer", and "The evidence on processed meat is even more clear-cut, with no level of intake associated with a lack of colorectal cancer risk."²¹ It gives a figure for red meat, "no more than about three portions per week", about 350 to 500 g cooked, and for processed meat, "Consume very little, if any, processed meat".²¹ It also names a mechanism: "Haem iron, which is present at high levels in red and processed meat, promotes colorectal tumorigenesis by stimulating the endogenous formation of carcinogenic N-nitroso compounds."²¹ (Tumorigenesis is tumour formation; endogenous means made inside the body.)

NutriRECS gave the advice in the opening paragraph. How it got there is the part worth following step by step.

How certainty becomes advice

Sleep lesson 6 taught that a guideline grade has two parts: the certainty of the evidence, and the strength of the recommendation, and that a weak or conditional recommendation turns on people's values and preferences. NutriRECS's guideline, read at abstract level, puts its method in one sentence: "Dietary guideline recommendations require consideration of the certainty in the evidence, the magnitude of potential benefits and harms, and explicit consideration of people's values and preferences."¹³ Take the three in turn, with the panel's own reviews.

Certainty. The cohort reviews rated the evidence low or very low. The reason, in one review's limitation section: "Inadequate adjustment for known confounders, residual confounding due to observational design, and recall bias associated with dietary measurement."¹⁴ Recall bias is error in what people remember eating, lesson 2's subject.

Magnitude. The panel's reviews expressed effects per three servings a week. For mortality and heart disease: "The magnitude of association between red and processed meat consumption and all-cause mortality and adverse cardiometabolic outcomes is very small, and the evidence is of low certainty."¹⁴ For cancer: "The possible absolute effects of red and processed meat consumption on cancer mortality and incidence are very small, and the certainty of evidence is low to very low."¹⁵ A third review, of dietary patterns, said low- or very-low-certainty evidence suggests less meat "may result in very small reductions" in harm.¹⁷

There's also a review of trials. "Of 12 eligible trials, a single trial enrolling 48 835 women provided the most credible, though still low-certainty, evidence that diets lower in red meat may have little or no effect on all-cause mortality (hazard ratio [HR], 0.99 [95% CI, 0.95 to 1.03])".¹⁶ You've met that trial, and lesson 2 read hazard ratios from it.

Values and preferences. The fifth review asked what people who eat meat want. Its conclusion: "Low-certainty evidence suggests that omnivores are attached to meat and are unwilling to change this behavior when faced with potentially undesirable health effects."¹⁸ It found "Thirteen studies reported that omnivores enjoy eating meat, 18 reported that these persons consider meat an essential component of a healthy diet".¹⁸

Put the three together and the panel's reasoning is visible. A very small possible benefit, at low certainty, set against something people value: under GRADE that gives a weak recommendation, and the panel's weak recommendation pointed to leaving intake where it was. The abstract adds what the panel left out on purpose: "Considerations of environmental impact or animal welfare did not bear on the recommendations."¹³ Those are value questions this course didn't research either, and it doesn't judge them.

WCRF and IARC reach their grades by another route. In this course's summary it runs through three questions: do many studies agree, does risk rise with the amount eaten, and is there a plausible way it could happen? For processed meat both answered yes. WCRF's page names haem iron and N-nitroso compounds as the way, IARC's red-meat grade rests partly on "strong mechanistic evidence", and the WHO page says the risk rises with the amount eaten.¹³²¹

The premise each needed

Now the step nobody wrote down. You know the evidence each body had, and what each said. The step between them is a premise, a general sentence about what matters, which neither states in so many words.

The case against NutriRECS, in its critics' words

The critics' main charges are about how the evidence was graded and how the numbers were presented.

Harvard's T.H. Chan School of Public Health answered on 30 September 2019 on its Nutrition Source site, a page this course read in full.¹⁹ Its central charge: "The new guidelines are not justified as they contradict the evidence generated from their own meta-analyses". On grading: "This should come as no surprise, since GRADE criteria were mainly developed for evaluating evidence from drug trials. Unlike drugs, dietary, lifestyle, and environmental factors are typically not amenable to large, long-term randomized clinical trials." On size: "First, the effect estimates may seem small because the unit of exposure (3 servings/week) is small." It reports that dietary patterns with a moderate reduction in red and processed meat "were associated with lower total mortality by 13% (95% Confidence interval 8% to 18%)", and lower risks of other outcomes. And on the panel: "these recommendations are not based on consensus of the panel because three panel members actually voted against their own recommendations."¹⁹

One of the reviews' own co-authors, John Sievenpiper, is quoted on the same page: "Unfortunately, the leadership of the paper chose to play up the low certainty of evidence by GRADE as opposed to the protective associations that directly support current recommendations to lower meat intake".¹⁹

In 2020 Qian, Riddle, Wylie-Rosett and Hu set out the grading argument in Diabetes Care; this course read the abstract and parts of the full text.²⁰ Their case is the one previewed above: the meat evidence should have been "upgraded", on grounds that include "dose-response, consistency across studies", mechanisms and trials on risk factors. They also point to NutriGrade, a scoring system built for nutrition, and to two earlier reviews that applied it, one rating the evidence linking red and processed meat to type 2 diabetes "high quality" and the other rating the evidence on mortality "moderate quality". NutriGrade's own paper, also read at abstract level, shows why it can rate cohorts higher: its score includes items for "effect size" and "dose-response", alongside risk of bias.²³ Qian and colleagues add a second objection: "inadequate attention was paid to what might be replacing red meat, be it plant-based proteins, refined carbohydrates, or other foods."²⁰ That's "instead of what?", pointed at the panel's advice to keep things as they are.

What the panel's side answers

In this course's summary, since it hasn't read a reply from the panel itself, the panel's case is in its method. Its reviews did not say there was no association; they reported one, in the expected direction, and called it very small and uncertain. Its limitation sections name the reasons, confounding and recall bias, which lessons 1 and 2 spent their time on. Its trial review found little or no effect in the one large trial there is. And it put the values step in the open: it ran a review of it and printed the result.

No trial is likely to settle it. Harvard's page says of a long trial of meat reduction that "no such trial has ever been conducted for practical and ethical reasons".¹⁹ So the argument is about how much weight cohorts can bear without one, which is lesson 1's argument between Ioannidis and the Harvard defenders, applied to one food.

An editorial supporting the reviews ran beside them; this course could not open it.

Who paid, and who has a stake

A conflict doesn't settle whether a finding is right. This course read what the abstracts print, one news report and one paper's disclosure statement, not the rest.

• NutriRECS. Each paper states "Primary funding source: None", and the guideline says "Strict criteria limited the conflicts of interest among panel members."¹³ A news report syndicated from the Washington Post, read in full, says that "In April, Johnston announced that the Agriculture and Life Sciences (AgriLife) program at Texas A&M would join the NutriRECS consortium and provide "generous support" to NutriRECS."²² Patrick Stover, AgriLife's dean and a guideline author, is reported there as saying the meat work was finished before the new funding.²² Bradley Johnston, the lead, was also senior author of the 2017 review of sugar guidelines that lesson 7 read, whose industry funding that paper disclosed. A later correction to the guideline, which this course saw only through search results, concerned an undisclosed grant; it did not read the correction.
• The critics. Johnston's reply, in the same report: "You might also look at the intellectual and financial conflicts of our adversaries."²² That's his stated defence. Qian and colleagues' paper discloses that Riddle has had research support and consulting fees from drug companies, and that Hu has had "research support from the California Walnut Commission" and lecture fees from Metagenics and Standard Process, "outside the submitted work".²⁰ This course did not read the Harvard page authors' disclosures or WCRF's funding. WCRF is a cancer charity whose own recommendation this is, a public stake of the kind Action on Salt had in lesson 7.

The journal's editor, Christine Laine, who had to decide whether to act, is quoted in the same report: "Would we retract the paper? No. A lot of the drug trials have industry funding. Conflicts of interest is just one potential source of bias."²² Whatever you make of her position, a conflict tells you where to look; the grading and the numbers are what you look at.

Separating the parts

So what, in the meat dispute, is fact, what is argued over, and what is a value?

Established, and agreed by every source here. Higher intakes of processed meat go with more bowel cancer, in the direction IARC, WCRF and NutriRECS's reviews all found.

Contested, and empirical. How certain that evidence is, which turns on whether consistent, dose-responsive cohort evidence with a mechanism can count as strong (WCRF, IARC, and NutriGrade in the reviews Qian and colleagues cite) or stays low (GRADE as NutriRECS applied it). What replaces the meat, to which, Qian and colleagues argue, the panel paid "inadequate attention". And how big the effect is: very small per three servings a week in NutriRECS's reviews, 13% lower mortality for a whole pattern in Harvard's reading of the same meta-analyses, and for bowel cancer about ten more cases per thousand heaviest eaters in Logic lesson 9's working. None of those units is dishonest.

Value. How to weigh a very small, uncertain reduction in risk against what people want to eat; whether advice should aim at the population or leave individual habits alone; and whether the environment or animals should count, which NutriRECS excluded and Qian and Hu's conclusion includes when it speaks of "human and planetary health".²⁰ This course describes those and does not decide them. Someone who agrees with every number can still hold either value premise.

As of September 2026 the US Dietary Guidelines for Americans, 2025–2030 say "Consume a variety of protein foods from animal sources, including eggs, poultry, seafood, and red meat, as well as a variety of plant-sourced protein foods", and tell readers to choose meat "with no or limited added sugars, refined carbohydrates or starches, or chemical additives". A search of the text found no explicit limit on processed meat. They tell readers to avoid "highly processed" foods without using NOVA's term.²⁴ Lesson 9 reads them.

What people get wrong

"Processed food is all the same." NOVA itself separates processed from ultra-processed, and Astrup's objection is that its fourth group is too mixed, taking in "several healthy, nutrient-dense foods".⁷

"The ultra-processed effect is just salt, sugar and fat." Hall's trays matched total salt, sugar and fat as served, and people still ate about 500 kcal a day more.⁹ But the kinds of sugar, fat and fibre, and the energy density of the solid food, still differed, so the trial rules out only the totals. The mirror is also wrong: "Hall proved processing itself is the harm." Its authors say it "was not designed to identify the cause", and energy density and eating rate are candidates.⁹

"A Group 1 carcinogen is as dangerous as tobacco." Sleep lesson 4 and Logic lesson 9 covered this: the group grades how strong the evidence is, not how big the risk.² The WHO page dated 23 September 2026 doesn't make the comparison at all.³

"NutriRECS showed meat is healthy," or "showed the cancer link is a myth." It suggested adults continue current intake, as a weak recommendation on low-certainty evidence, and its own reviews found small associations in the same direction as everyone else's.¹³¹⁴¹⁵

"NutriRECS had industry ties, so its numbers are wrong." The disclosure problems are documented.²² But Harvard's case against it was built on "the evidence generated from their own meta-analyses", and the dispute is over grading and meaning, not the arithmetic.¹⁹

Practice

Connections

Back. Lesson 1's argument between cohorts and trials is the grading dispute over meat, and its "instead of what?" is Qian and Hu's second objection. Lesson 2's Women's Health Initiative was the one large trial in NutriRECS's review, and lesson 7 met Johnston grading the sugar guidelines. Sleep lesson 6 taught the two parts of a guideline grade.

Forward. Lesson 9 reads whole dietary patterns and the national guidelines built on them, including the US guidelines quoted above.

Go deeper

• Hall and colleagues, 2019, Cell Metabolism, free at PubMed Central. Read here: the abstract and, from the full text, the design, the main results, the limitations and the conclusion. Its "Limitations of Study" section is a model of saying what a trial can't show.
• The 2022 AJCN debate, Monteiro for and Astrup against. Read here: both abstracts.
• Johnston and colleagues, 2019, the NutriRECS guideline, in the Annals of Internal Medicine. Read here: the abstract only. Read it beside its values review to see the step from evidence to advice written down.
• Harvard's reply, 30 September 2019. Read here: the whole page. The critics' case in one place, co-author's dissent included.

Sources

1. International Agency for Research on Cancer, Press Release No. 240, 26 October 2015, PDF. Read: the whole PDF. The underlying Lancet Oncology summary (Bouvard 2015) was not read.
2. IARC, Monographs Q&A on red and processed meat, volume 114, 2015, PDF. Read: the whole PDF.
3. World Health Organization, questions and answers on red and processed meat and cancer, page dated 23 September 2026. Read: the whole page, on 24 September 2026. What it doesn't carry, compared with IARC's 2015 Q&A, is from a search of the page text; no earlier version of this page was read.
4. C. A. Monteiro and colleagues, "Ultra-processed foods: what they are and how to identify them", Public Health Nutrition 22(5), 2019, doi 10.1017/S1368980018003762. Read: the abstract. The critics' purpose objection and the reading of the test as Monteiro's answer are this course's summaries.
5. M. J. Gibney, C. G. Forde, D. Mullally and E. R. Gibney, "Ultra-processed foods in human health: a critical appraisal", American Journal of Clinical Nutrition 106(3), 2017, doi 10.3945/ajcn.117.160440. Read: the abstract.
6. C. A. Monteiro and A. Astrup, "Does the concept of "ultra-processed foods" help inform dietary guidelines, beyond conventional classification systems? YES", AJCN 116(6), 2022, doi 10.1093/ajcn/nqac122. Read: the abstract.
7. A. Astrup and C. A. Monteiro, the same question, "NO", AJCN 116(6), 2022, doi 10.1093/ajcn/nqac123. Read: the abstract. The gloss on glycemic load is this course's.
8. V. Braesco and colleagues, "Ultra-processed foods: how functional is the NOVA system?", European Journal of Clinical Nutrition 76(9), 2022, doi 10.1038/s41430-022-01099-1. Read: the abstract. "Three quarters" is this course's arithmetic from 90 of 120.
9. K. D. Hall and colleagues, "Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake", Cell Metabolism 30(1), 2019, doi 10.1016/j.cmet.2019.05.008. Read: the abstract, and from the full text the introduction, the diet design paragraph, the results on energy intake, energy density, eating rate, appetite ratings, hormones and glucose tolerance, "Limitations of Study", and the conclusion. The glosses on standard error and on the protein target are this course's.
10. S. J. Dicken and colleagues, "Ultraprocessed or minimally processed diets following healthy dietary guidelines on weight and cardiometabolic health: a randomized, crossover trial", Nature Medicine 31(10), 2025, doi 10.1038/s41591-025-03842-0. Read: the abstract only. The reading of the weight loss on both diets is this course's.
11. M. M. Lane and colleagues, "Ultra-processed food exposure and adverse health outcomes: umbrella review of epidemiological meta-analyses", BMJ 384, 2024, e077310, doi 10.1136/bmj-2023-077310. Read: the abstract. The explanation of how one finding gets both grades is this course's reading; the abstract doesn't define its classes.
12. C. A. Monteiro and colleagues, "Ultra-processed foods and human health: the main thesis and the evidence", Lancet 406(10520), 2025, doi 10.1016/S0140-6736(25)01565-X; and G. Scrinis and colleagues, "Policies to halt and reverse the rise in ultra-processed food production, marketing, and consumption", Lancet 406(10520), 2025, doi 10.1016/S0140-6736(25)01566-1. Read: both abstracts. The 2026 letters in reply were not read.
13. B. C. Johnston and colleagues, "Unprocessed Red Meat and Processed Meat Consumption: Dietary Guideline Recommendations From the Nutritional Recommendations (NutriRECS) Consortium", Annals of Internal Medicine 171(10), 2019, doi 10.7326/M19-1621. Read: the abstract, which notes that the article has been corrected. The full text was blocked.
14. D. Zeraatkar and colleagues, "Red and Processed Meat Consumption and Risk for All-Cause Mortality and Cardiometabolic Outcomes", Annals 171(10), 2019, doi 10.7326/M19-0655. Read: the abstract.
15. M. A. Han and colleagues, "Reduction of Red and Processed Meat Intake and Cancer Mortality and Incidence", Annals 171(10), 2019, doi 10.7326/M19-0699. Read: the abstract.
16. D. Zeraatkar and colleagues, "Effect of Lower Versus Higher Red Meat Intake on Cardiometabolic and Cancer Outcomes: A Systematic Review of Randomized Trials", Annals 171(10), 2019, doi 10.7326/M19-0622. Read: the abstract. That its single large trial is the Women's Health Initiative comes from Harvard's page, not the abstract.
17. R. W. M. Vernooij and colleagues, "Patterns of Red and Processed Meat Consumption and Risk for Cardiometabolic and Cancer Outcomes", Annals 171(10), 2019, doi 10.7326/M19-1583. Read: the abstract.
18. C. Valli and colleagues, "Health-Related Values and Preferences Regarding Meat Consumption: A Mixed-Methods Systematic Review", Annals 171(10), 2019, doi 10.7326/M19-1326. Read: the abstract.
19. Harvard T.H. Chan School of Public Health, The Nutrition Source, "New “guidelines” say continue red meat consumption habits, but recommendations contradict evidence", 30 September 2019, with a correction note of 2 October 2019. Read: the whole page.
20. F. Qian, M. C. Riddle, J. Wylie-Rosett and F. B. Hu, "Red and Processed Meats and Health Risks: How Strong Is the Evidence?", Diabetes Care 43(2), 2020, doi 10.2337/dci19-0063. Read: the abstract, and from the full text the NutriGRADE section, Table 1, the "Duality of Interest" statement and references 5, 33 and 34. The two NutriGrade ratings are those references' work, which the paper cites.
21. World Cancer Research Fund, "Limit consumption of red and processed meat", Cancer Prevention Recommendations evidence page. Read: the whole page. No date visible on it.
22. "Research group that discounted risks of red meat has ties to program backed by beef industry", Washington Post story syndicated in the Grand Forks Herald, late 2019 (undated on the text read). Read: in full. The 31 December 2019 correction in the Annals (doi 10.7326/L19-0822) was seen only in search results and not read.
23. L. Schwingshackl and colleagues, "Perspective: NutriGrade: A Scoring System to Assess and Judge the Meta-Evidence of Randomized Controlled Trials and Cohort Studies in Nutrition Research", Advances in Nutrition 7(6), 2016, doi 10.3945/an.116.013052. Read: the abstract.
24. US Departments of Health and Human Services and of Agriculture, Dietary Guidelines for Americans, 2025–2030, January 2026, PDF. Read: the whole document. Supports the dated line only. The search of its text for a processed-meat limit was done for this course.